Kuberntes OOM Error. Very High memory usage > 15gb

[josekidengan]

Kubernetes memory usage is spiking more than 15Gb, but local run on ubuntu takes much less. I figured out that Kubernetes does not use swaps. However, I could not find any specific parameter to control this spiking of memory in case a swap is not present.
Even At 2 threads, memory usage spikes to ~15GB.

[HenryCaiHaiying]

You can use 'jmap -histo' to print out the number of objects per class, run it a few times to see which class of objects keep increasing.

…

On Tue, Jun 29, 2021 at 9:21 AM Jose Thomas ***@***.***> wrote: Although secor is working, it looks like the memory usage is constantly increasing over time eventually shuts off the system. All I could sense is an info (org.apache.hadoop.io.compress.CodecPool) INFO Got brand-new compressor [.gz] It looks like the object is being created without recycling and memory leaks is causing the system to shut down. Has anyone got a similar issue Related config secor.compression.codec=org.apache.hadoop.io.compress.GzipCodec — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#2129>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABYJP77RSLDRSRNRNRTKSNLTVHXILANCNFSM47QQAVHA> .

[josekidengan]

You can use 'jmap -histo' to print out the number of objects per class, run it a few times to see which class of objects keep increasing.
…
On Tue, Jun 29, 2021 at 9:21 AM Jose Thomas @.***> wrote: Although secor is working, it looks like the memory usage is constantly increasing over time eventually shuts off the system. All I could sense is an info (org.apache.hadoop.io.compress.CodecPool) INFO Got brand-new compressor [.gz] It looks like the object is being created without recycling and memory leaks is causing the system to shut down. Has anyone got a similar issue Related config secor.compression.codec=org.apache.hadoop.io.compress.GzipCodec — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#2129>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABYJP77RSLDRSRNRNRTKSNLTVHXILANCNFSM47QQAVHA .

I update the comments and description with my insights.